Tommy Tuberville PFE: a high composite, a low ML head, and the question they don't both answer

Headline correlation

Tommy Tuberville (R-Alabama), PFE (Pfizer Inc.)

Composite score: 81.6 / 100 (high tier)


Transaction	sale_full, $15,001 - $50,000, 2026-06-08
Disclosure	2026-07-16 (38 day lag)
Sector	Health Care to Pharmaceuticals
Committees	Senate Committee on Health, Education, Labor, and Pensions, Senate Committee on Veterans’ Affairs, Senate Special Committee on Aging
ML annotation	ML heads diverge: distill 100.0, forward-CAR 29.1. The two model estimates disagree, and the gap is the story; see methodology. Both are model scores, not realized returns. Measured abnormal returns vs SPY are reported separately in the returns row.

A note on the label: 2026-W24 is the ISO week of the headline trade itself. Each issue covers correlations from the trailing week of disclosures, which can surface trades executed weeks earlier.

What the engine saw. The trade involving Tommy Tuberville’s sale of Pfizer shares was flagged due to a high correlation score of 81.57/100, driven by event timing (trade occurred 0 days apart from relevant events), committee overlap in the healthcare sector, and a pre-event trade pattern. These factors, combined with lobbying activities by healthcare entities targeting CMS and HHS, suggest potential alignment between the trade and policy-related events, though no definitive intent or wrongdoing is implied.

Supporting signals.

Bill (score 81.6, 2026-06-08): Right to Try for Individualized Treatments Act
Bill (score 63.9, 2026-06-24): Stop the Sexualization of Children Act
Bill (score 59.4, 2026-06-04): Homeopathic Drug Product Safety, Quality, and Transparency Act

SHAP top features. timing_raw (+7.954), disclosure_lag_raw (+6.396), jurisdiction_raw (+4.902)

Abnormal returns vs SPY. n/a (7d) / -7.2% (30d) / n/a (90d)

Why this one matters

Start with the “diverged” label, because this week it is doing more work than usual. The forward-CAR head is trained on the stock’s abnormal return after the trade date, with no adjustment for which side of the trade the member was on. Its low score (29.1) is a prediction that Pfizer would underperform the market in the weeks after June 8, and Pfizer did: the sidecar shows a 30-day abnormal return of -7.2% against SPY. Read purchase-side, that is “the trade did not pay off.” But this was a full exit. A seller who gets out ahead of a seven-point relative slide made the right call by the market’s own scorecard. For a sale, the two heads are not actually arguing; a high-scoring correlation and a well-timed exit can both be true, and here they appear to be.

Now the counterweight, because it is substantial. The Pfizer sale was not a pharma call in isolation. The same July 16 filing disclosed eleven sales, ten of them on June 8: CSX, Lockheed Martin, Duke Energy, NextEra, American Water Works, Mastercard, Procter and Gamble, Accenture, Tractor Supply, and Pfizer, with Wabtec following on June 9. These were Tuberville’s first disclosed trades of 2026, made in the middle of his run for governor of Alabama. A same-day, cross-sector sweep reads like a portfolio liquidation, not a committee-informed pick. The timing factor that tops the SHAP attributions is thinner than the score suggests, too: the June 8 “event” is S.4698, the Right to Try for Individualized Treatments Act, introduced and referred to his Health, Education, Labor, and Pensions Committee that day. It is a niche bill about patient-specific therapies, not something that moves a mega-cap pharma price.

So the honest read is: the composite did its job (a HELP member exited a pharmaceutical position the day a pharma bill hit his committee’s docket, disclosed it 38 days later near the STOCK Act’s 45-day limit, and the stock then underperformed), and the surrounding context supplies a plausible innocent explanation the engine cannot see. Tuberville has traded some $38 million in securities since 2021, opposes a congressional trading ban, and says brokers manage his account without his input. If that is true, the timing belongs to a broker cleaning up a portfolio ahead of a statewide campaign. What would change our mind: selectivity. If later filings show health-care positions were exited while comparable holdings elsewhere were kept, or if the exits cluster around HELP calendar items with real pricing consequences (the drug-pricing docket, not right-to-try), this stops looking like housekeeping. Until then, this is a correlation worth logging and watching, not a smoking gun.

Sources

Facts in this section that come from outside the engine’s dataset:

Full list of the June 8-9 sales, the $38 million traded since 2021, and Tuberville’s public statements on broker-managed accounts and the trading ban: Benzinga, July 17, 2026
S.4698, Right to Try for Individualized Treatments Act: introduced June 8, 2026 and referred to the Senate HELP Committee the same day: Congress.gov
The trade, disclosure, committee, and abnormal-return figures are from this brief’s data sidecar; the 30-day abnormal return is computed against SPY per the methodology.

Runner-ups this week

Daniel Crenshaw (R) to AAPL (Information Technology): score 80.9, traded 2026-06-01
Daniel Crenshaw (R) to META (Communication Services): score 77.7, traded 2026-06-01
Daniel Crenshaw (R) to GOOG (Communication Services): score 77.7, traded 2026-06-01
Thomas H. Kean Jr (R) to JNJ (Health Care): score 70.3, traded 2026-06-29
Daniel Crenshaw (R) to META (Communication Services): score 65.0, traded 2026-05-05

Newcomers we’re watching

Members with under 5 total trade disclosures get listed but not scored; we want enough history to compare against before reading signal into their patterns. This week’s newcomers we’re watching:

Debbie Wasserman Schultz (D): 1 lifetime trade(s); most recent: ICHR on 2026-06-17
Robert J. Wittman (R): 1 lifetime trade(s); most recent: CCI on 2026-06-30
Greg Stanton (D): 2 lifetime trade(s); most recent: TCNNF on 2026-05-28
Susie Lee (D): 3 lifetime trade(s); most recent: FLL on 2026-07-01
Debbie Dingell (D): 3 lifetime trade(s); most recent: HONAV on 2026-06-29

This week in the dataset

85 new trade disclosures ingested
2,264 new bill records
12,555 new lobbying filings
4,956 new correlations scored
Top composite score this week: 81.6

Source: public records (QuiverQuant trades, USAspending contracts, Congress.gov bills, lda.gov lobbying). Published under CC0. See the full methodology for the scoring formula, and the supporting data for replication inputs.